计算机应用 ›› 2016, Vol. 36 ›› Issue (4): 905-908.DOI: 10.11772/j.issn.1001-9081.2016.04.0905

• 网络与通信 • 上一篇    下一篇

基于故障传播模型与监督学习的电力通信网络故障定位

赵灿明1, 李祝红1, 陶磊2, 张信明2   

  1. 1. 国家电网 芜湖供电公司, 安徽 芜湖 241000;
    2. 中国科学技术大学 计算机科学与技术学院, 合肥 230027
  • 收稿日期:2015-10-02 修回日期:2015-12-04 出版日期:2016-04-10 发布日期:2016-04-08
  • 通讯作者: 张信明
  • 作者简介:赵灿明(1983-),男,安徽太湖人,工程师,硕士,主要研究方向:智能电网、电力信息网络; 李祝红(1974-),男,安徽怀宁人,高级工程师,硕士,主要研究方向:智能电网、电力信息网络; 陶磊(1992-),男,安徽和县人,博士研究生,主要研究方向:无线网络、智能电网;张信明(1964-),男,安徽天长人,教授,博士,CCF高级会员,主要研究方向:无线网络、智能电网。
  • 基金资助:
    国家自然科学基金资助项目(61379130)。

Fault localization for electric power communication network based on fault propagation model and supervised learning

ZHAO Canming1, LI Zhuhong1, TAO Lei2, ZHANG Xinming2   

  1. 1. Wuhu Power Supply Company, State Grid, Wuhu Anhui 241000, China;
    2. School of Computer Science and Technology, University of Science and Technology of China, Hefei Anhui 230027, China
  • Received:2015-10-02 Revised:2015-12-04 Online:2016-04-10 Published:2016-04-08
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61379130).

摘要: 针对电力通信网络中的故障定位问题,分析了一种网络设备或链路故障引发的大范围连通片故障告警情形,提出一种基于故障传播模型和监督分类学习方法的故障定位算法。首先使用改进的故障传播模型求得初步定位结果,用最少的故障数目解释当前告警;然后通过故障源-故障告警向量分解将故障定位问题转化为监督分类问题,定位告警区域内部故障;最后加入猜测的故障设备和故障链路完善定位结果以提高定位准确率。模拟结果表明提出的故障定位算法的故障检测率达到84%~95%,具有较高的故障定位可靠性。

关键词: 故障定位, 监督学习, 故障传播模型, 电力通信网络

Abstract: To solve the fault localization problem in electric power communication network, the large-scale connected area fault alarms caused by device or link faults were investigated, and a fault localization algorithm based on fault propagation model and supervised learning method was proposed. First, an improved fault propagation model was used to obtain an initial result with the minimum faults. Then the fault localization problem was transformed into a supervised classification problem by fault alarm vector decomposition to localize the faults within the fault warning areas. Finally, conjectural fault devices and links were added to improve the location results of previous two steps and increase the accuracy. The simulation results show that accuracy of fault localization of the proposed algorithm reaches 84%-95%, which achieves high reliability in fault location.

Key words: fault localization, supervised learning, fault propagation model, electric power communication network

中图分类号: